Editor’s note: Dan Lockhart, Ph.D., is director of marketing sciences, and Fred Detwiler, Ph.D., is a research analyst, at Maritz Marketing Research Inc., St. Louis.

The Internet is growing both in its number of users and as a medium for commerce and advertising. CyberAtlas 1, an on-line statistical reference resource for Internet marketing, reports that there are a total of 40-45 million adults in the U.S. using the Internet and 25 million total weekly Web users. They further report that some companies are netting on-line sales totals of up to $2 million per day, and they estimate second quarter Internet advertising revenue at $133 million.

This article discusses the following information regarding the demographics of individuals on the Internet:

  • number of people on the Internet;
  • current picture of Internet sales and advertising;
  • information on companies who are conducting research on or about the Internet;
  • gender, age, and income biases associated with the Internet;
  • interpretation of the meaning of results comparing Internet and telephone surveys.

Market size

Estimates of the number of people on the Internet vary depending on which of the following qualifications one uses:

  • daily use of the Internet;
  • having an e-mail account;
  • connections in the last three months;
  • ever looked at a computer screen connected to the Internet.

In addition, various groups have reasons to overstate or understate the actual market size of the Internet audience. While CyberAtlas reports that there are 25 million total weekly Web users and more than 31 million computers regularly connected to the Internet, forecasts from FIND/SVP 2, a provider of worldwide research and business intelligence services, indicate that by year’s end there will be:

  • 75 million adult users in the U.S.;
  • 28 million children;
  • 28 million Internet-using households.

Peter Krasilovsky 3 reported in early 1996, "Twenty percent of the American workforce and 5 percent of U.S. households say they currently have access to the Web…For the most part, these users represent the small (10 percent) but influential, highly-educated, high income, ‘early adopter’ segment of the population."

Sales

USA Today 4 reports that salaries for a "vice president, on-line" are in the six figures. Their findings also indicate that Internet sales generate profits for 15 of the 104 large companies it surveyed. Web Commerce 5 reports that The Yankee Group indicates that business-to-business electronic commerce will exceed $135 billion by the year 2000. They further indicate that manufacturers of computer and airplane parts, such as Cisco and Boeing, lead the pack in business-to-business Web commerce this year, with $3 billion in sales, or 38 percent of Internet business trade 6.

Dell Computer’s "Internet-Direct" reports that its Net-based sales are growing at a rate of 20 percent per month and generating revenues of more than $1 million a day. Dell reportedly has corporate customers purchasing $30,000 high-end servers on-line, and one large Fortune 500 company estimates its annual savings to be $2 million through its Dell-developed Web site. Cox News Service 7 reports that sales of music-related merchandise over the Internet now accounts for approximately $23 million yearly.

Advertising

According to CyberAtlas, second-quarter Internet ad revenues totaled $133 million - a 58 percent increase over the first quarter’s $84 million and a 250 percent increase over the same time period in 1996. Some heavy-hitting on-line advertisers include Microsoft, TV Guide, and Wired magazine.

However, CNET 8, a digital technology information source, reports that the companies behind major consumer brands are still reluctant to advertise on the Internet. While computer giants are swarming at advertising on the Internet, consumer products are holding back. Peter Storck of Jupiter Communications 9 claims that the large consumer goods providers want solid evidence comparing various methods of advertising before committing. CNET also reports that computer and Internet related companies spent 54 percent of their advertising budget for on-line advertising. Current Internet ad leaders include Netscape Communications, Yahoo!, Infoseek, and Excite.

CNET shows that automotive companies make up 5 percent of Internet advertising presence, with other consumer companies comprising 15 percent of on-line ads. Jupiter Communications expects on-line advertising spending to exceed $5 billion by the year 2000.

Internet surveys and research

Several Internet sites are reporting on the on-line surveys they have conducted. Some of these are:

  • Georgia Tech’s historical Internet surveys;
  • O’Reilly & Associates RDD telephone survey of Internet users;
  • Netcraft’s survey of Web servers;
  • Find/SVP’s American Internet User survey;
  • Maritz AmeriPoll On-line;
  • Nielsen’s Internet demographics survey;
  • sri VALS 2 Internet survey.

These surveys generally show that the Internet is gradually becoming more representative of the general American population. Some of the results of recent surveys are shown in Table 1.

Georgia Tech indicates that the average age has slowly but steadily been increasing, and the gender bias has been decreasing. Table 2 shows the decrease in gender bias as the percentage of female users steadily increases across the seven surveys conducted by Georgia Tech.

Age bias

Past studies have shown that Internet samples have an age bias, with samples tending to overrepresent the middle-age categories and underrepresent the oldest age categories. Table 3 compares U.S. Census data to surveys by Find/SVP, O’Reilly & Associates, and Maritz AmeriPoll On-line. The Find/SVP and Maritz age categories were adjusted to match those of the other surveys.

Sri 10 reports that there is a change in the age demographics of the Internet depending on the time of year when a survey is conducted. "We measured a sharp decline in age after mid-August 1995, when the fall academic year started. As the domain data show, our sample contained a significantly larger proportion of new and returning university students from August through the start of October."

Table 4 shows the slight change in the average age of respondents to Georgia Tech’s surveys.

Georgia Tech notes that the average age has not changed much over time, but distribution has. The Internet population used to have a greater peak in the middle age bracket. Over time, a greater number of older and younger users are coming on-line, causing the age curve to become flatter.

Income bias

The data in the chart from the Georgia Tech studies (below Table 6) indicates that individuals on the Internet are more likely to be in the wealthiest category than are individuals in the census.

However, data comparing previous surveys by Georgia Tech indicate that there has been a gradual decline in the average income of individuals surveyed in the Georgia Tech interviews (see Table 6).

Baseline knowledge regarding research on the Internet is rapidly increasing. Past surveys have suggested differences in samples obtained on the Internet versus samples obtained via other methods. However, there has been little attempt to validate the differences observed. At face value, it seems logical that individuals on the Internet are more likely to be younger males with more income than individuals not on the Internet. In addition, it seems to make sense that individuals on the Internet would be more likely to own (or have access to) a computer than individuals not on the Internet.

But beyond this superficial validity, little is known with much precision. Georgia Tech has conducted a number of surveys and compared its results to those found by Find/SVP. However, if both surveys are biased, there is no definitive baseline of comparison. Therefore, if someone obtains a different result than these sources, it is not clear which source is the most accurate. In addition, other factors may account for differences observed, such as:

  • dissimulation or lying;
  • misunderstanding the questions;
  • differences in the interpretation of questions.

Maritz is beginning to explore differences in survey results obtained via the Internet and those obtained via other methods. Currently, we have completed five parallel studies via both the Internet and telephone. Table 7 shows some basic information on these five studies.

Preliminary results of our analyses indicate that there are differences in the responses provided by individuals in an Internet survey and those provided by individuals in a telephone survey. Our data further indicate that these differences are not accounted for by differences in the demographics of respondents. That is, when one takes into account differences in gender, age, and sex, there are still discrepancies in responses not accounted for. Analysis of 18 variables from three of the Maritz studies (billboards, car washing, and holiday shopping I) reveals these telephone vs. Internet differences. Table 8 shows the eight variables on which the respondents differed.

It should be emphasized that these studies only examined telephone vs. Internet responses. The fact that some phone respondents are Internet users makes it likely that the differences found are smaller than would be expected if Internet users had been screened out of the telephone responses. The eight significant relationships out of 18 tests are higher than would be expected by chance alone. These data show that Internet responses are different in the following manner:

  • They feel less safe in their cars.
  • They find billboards more entertaining.
  • They are less likely to believe that billboards should be regulated.
  • They wash their cars less frequently.
  • They change their oil less frequently.
  • They plan further ahead for holiday shopping.
  • They begin making their holiday purchases further ahead.
  • They are more annoyed by pre-Halloween holiday sales.

Going beyond these results leads us to believe that individuals on the Internet are less conscientious about their cars, plan further ahead, are more entertained by advertisements, and are more interested in information and freedom of information flow. One may see an apparent contradiction between these individuals planning ahead yet being less conscientious about their cars. This could be accounted for by a belief by these individuals that washing one’s car and changing the oil (slightly less often) does not have a significant impact on their automobiles. These individuals may be more concerned about self-improvement, knowledge, and civil liberties, and less concerned about their material possessions.

A recent presentation by Roberta McConochie 11 at TECH ’98 (sponsored by Golden Gate University, MRA, CASRO, Silicon Valley AMA, and the San Francisco Bay Area AMA) adds further insight into these results. She has found that individuals on the Internet are more likely to fit into one of three Arbitron segments: fast-laners, savvy sophisticates, or sports fanatics. She indicates that these three segments share the following characteristics:

  • They prefer fast-paced entertainment.
  • They use technology to attain goals.
  • They are achievement oriented.
  • They are more involved in sports.
  • They tend to have higher incomes.
  • They are more likely to be college graduates.
  • They are more entertained by fast-paced, intense, interactive entertainment.

These individuals appear to be information seekers who tend to be innovators or early-adopters in their purchase behavior. Over time, more and more laggards will be getting on the Internet. We believe that because the Internet is a more private place for interviews and represents an environment that is more conducive to honest responses, social dissimulation will be less for Internet surveys than for telephone surveys.

The Internet represents a "brave new world" for professional marketing research organizations, where on-line surveying standards are just starting to be discussed. Currently, there is no definitive source identifying the population of individuals on the Internet. As this article has shown, the evidence indicates that the Internet has been very biased toward males with higher incomes in middle-to-younger age brackets. Over time, these biases seem to be lessening.

The Internet is already a vast medium for commerce and information exchange. Profitable companies are advertising and selling their goods on the Internet. New methods to legally and illegally make money are being made possible by the Internet. Our studies work toward helping to define how research should be conducted on the Internet, what differences one can expect when doing research on the Internet, and how research can be used to better understand the Internet.

Notes

1. CyberAtlas (1997). [On-line] Available: http://www.cyberatlas.com/marlet/html

2. Find/SVP [On-line] Available: http://etrg.findsvp.com/internet/overview.html [August 1, 1997].

3. Peter Krasilovsky, "Marketing on the Information Superhighway," Marketing Tools, Jan/Feb 1996, pp. 48-49.

4. Jones, Del, USA Today (1997). [On-line] Available: http://www.news.com/News/Iten/ 0,4,11356,00html

5. Web Commerce. [On-line] Available: http://pubs.cmpnet.com/cw/webcommerce/659web.htm [May 8, 1998].

6. Web Commerce. [On-line] Available: http://pubs.cmpnet.com/cw/webcommerce/web0804 1.html [May 8, 1998].

7. Cox News Service (1997), "Cyberspace Music Rockin’;" St. Louis Post-Dispatch, Aug. 6, 1997, p. 7C.

8. CNET (1997). [On-line] Available: http://www.news.com/News/Item/0,4,5905.html

9. Jupiter Communications (1997). [On-line] Available: http://www.jup.com/

10. sri (1997). [On-line] Available: http://future.sri. com/vals/trends

11. Roberta McConochie, "Distinctive Web Sub-populations: Three of Eight Pathfinder Segments" (Arbitron NewsMedia) at TECH ’98, San Francisco.